Using neural networks in trading - page 40

New comment
 
I would first focus on the learning function - what to minimise and how, and whether or not to change the functional in the learning process. Then I would focus on the learning method (I don't want genetic, I want something fast and clear like Levenberg-McWardt). Then I want to estimate computational complexity. And if it turns out to be unbearable, I would switch to paralleling.
 
Think about it for now. And yet, it is better to think about paralleling as the problem is being formulated. It may affect the formulation and choice of algorithms, which is good and right.
 
Lycos:

Hello all!

I have read the information on nets and understood that they can predict everything, but I just need to know what.

In any business, you need to know the HOW, not the WHAT. The WHAT is found out as the play unfolds.
alsu:

I read such an entertaining article.

I am interested in the principle of brain structuring as a neural network - at first a large number of all kinds of connections are accumulated without paying much attention to their quality, and then selection is started by the principle "remove everything that is not necessary to achieve the result".

Well, yes. In order to make artificial intelligence work a bit, you have to work a lot with natural intelligence.

Good video on the subject:

1...3334353637383940
New comment